lable at ScienceDirect

Applied Geography 56 (2015) 145e153

Contents lists avai

Applied Geography

journal homepage: www.elsevier .com/locate/apgeog

The role of local land-use on the urban heat island effect of Tel Aviv asassessed from satellite remote sensing

Orit Rotem-Mindali*, Yaron Michael, David Helman, Itamar M. LenskyDepartment of Geography and Environment, Bar Ilan University, Ramat Gan 5290002, Israel

a r t i c l e i n f o

Article history:Available online

Keywords:MODISNDVISurface temperatureUHIUrban planningVegetation

* Corresponding author. Tel.: þ972 3 5318614; fax:E-mail address: orit.rotem@biu.ac.il (O. Rotem-Min

http://dx.doi.org/10.1016/j.apgeog.2014.11.0230143-6228/© 2014 Elsevier Ltd. All rights reserved.

a b s t r a c t

Climate change in cities has received much focus in the past few decades. Heat stress in urban areas hasan adverse effect on human health and is expected to worsen in the future due to the global warming.Vegetation has been shown to mitigate this effect, but introducing ‘green’ areas into the metropolitanspace is a challenging task. We assessed the thermal load in terms of surface temperature in Tel Aviv, thebiggest metropolitan area of Israel. The thermal effect of four different urban land uses was estimated.Specifically, we compared the cooling effect of residential areas with high vegetation cover (referred hereas ‘green’ residential) to that of small to medium size (2e40 ha) public parks. To this end, we usedsatellite data of land surface temperature (LST) and the Normalized Difference Vegetation Index (NDVI),as a surrogate for vegetation cover. High-temporal data were combined with high spatial resolutions datato produce 10-year average LST and NDVI maps at high spatial resolution over Tel Aviv. As expected,industrial areas had the highest LST due to lowest ratio of vegetation to free space area (1%), while ‘green’areas displayed the lowest LST. Green residential and small-medium public parks had comparablethermal loads, with green residential having slightly lower LST (by 0.5 �C). In general, small-mediumpublic parks displayed higher LST than expected. Inefficient use of free spaces for vegetation, i.e., rela-tively low vegetation cover to free space ratio, was probably the main cause for this. Public parks had ahigher local cooling effect, but a less continuous one on the proximate surrounding (30e90 m from thepark), probably due to their relative location in the urban fabric. Our results suggest that ‘greening’ areaswithin the private urban space should be encouraged at the expense of building new small-mediumparks in metropolitan areas that lack the sufficient free space for larger parks. The outcome of thisstudy may have key implications for urban planners seeking to mitigate urban heat island effects underthe limitation of existing dense urban layout.

© 2014 Elsevier Ltd. All rights reserved.

Introduction

Climate change in cities has received much focus in the past fewdecades. Urban heat islands (UHI), inwhich urban areas arewarmerthan their surroundings, is caused by altered surface cover andanthropogenic activity (Kleerekoper, van Esch, & Salcedo, 2012;Oke, 1995). Heat stress in cities is expected to worsen in thefuture due to the adverse effect of increasing urbanization andglobal warming on human health (Johnson, Stanforth, Lulla, &Luber, 2012; Kovats & Hajat, 2008). Numerous studies have pro-posed strategies to mitigate this effect. Most studies suggestincreasing the ‘green’ urban areas, since plants have a cooling effect

þ972 3 7384033.dali).

through transpiration and shade (Bowler, Buyung-Ali, Knight, &Pullin, 2010; Muller, Kuttler, & Barlag, 2013; Oke, Crowther,McNaughton, Monteith,& Gardiner, 1989; Potchter, Cohen,& Bitan,2006; Rogan et al. 2013; Spronken-Smith& Oke,1998). However, asmetropolitan areas usually lack the required public spaces,greening of building façades and street tree planting in residentialneighborhoods could be the most practicable solution (Gill,Handley, Ennos, & Pauleit, 2007).

Urban planners have understood that a knowledgeable design ofthe city is essential to reduce the burden of heat stress (Stone, Hess,& Frumkin, 2010). An increase of 2 �C to 5 �C is expected due tofuture expansion of urban areas (Lin, Liu, Ma, Li,& Shi, 2013). A builtenvironment changes the absorption and reflection of solar radia-tion and the surface energy balance (Bowler et al. 2010). Anthro-pogenic factors that contribute to this effect are fossil fuelconsumption (e.g., vehicles and air conditioning), construction

O. Rotem-Mindali et al. / Applied Geography 56 (2015) 145e153146

materials (e.g., stones and concrete vs. wood), open spaces (pavedroads and parking lots vs. green parks and trees), and citymorphology (e.g., high buildings vs. private houses) (Mirzaei &Haghighat, 2010; Petralli, Massetti, Brandani, & Orlandini, 2014).All of these are referred to as urban design and are regarded asurban design indicators.

Although there are several strategies to mitigate the UHI effect,most studies focus on the cooling effect of urban parks and gardens(Bowler et al. 2010; Cohen, Potchter, & Matzarakis, 2012; Eliassonand Upmanis 2000; Feyisa, Dons, & Meilby, 2014; Kleerekoperet al. 2012; Shashua-Bar & Hoffman, 2000). For example,Shashua-Bar and Hoffman (2000) showed that urban parks mayhave a cooling effect of up to 1.5 �Ce3 �C, depending on the time ofday. Oliveira, Andrade, and Vaz (2011) concluded that even smallparks of ~0.24 ha could mitigate the adverse effects of UHI and thepotential additional effects of global warming in cities. Othersshowed that medium to large size parks (>150 ha) have an evengreater thermal effect (Alcoforado, Andrade, Lopes, & Vasconcelos,2009; Andrade & Vieira, 2007; Eliasson and Upmanis 2000;Jauregui, 1990; Shashua-Bar & Hoffman, 2000). In Taipei andManchester, large parks have important cooling effects, implyingthat vast green spaces constitute critical environmental capital tomitigate climate effects (Chang, Li, & Chang, 2007; Gill et al. 2007).

An important aspect of the park cooling effects is their thermalinfluence on the immediate surrounding area. It was shown thattemperatures gradually increased with increasing distance fromthe park, up to approximately 500 me1000 m (Papangelis,Tombrou, Dandou, & Kontos, 2012; Ren et al. 2013; Upmanis,Eliasson, & Lindqvist, 1998; Yu & Hien, 2006). This spatial coolingeffect was found to be related to the park size, with a greater effectfor larger parks up to a certain threshold (Ren et al. 2013).Conversely, Feyisa et al. (2014) reported a maximum park coolingdistance of only 240 m and a maximum park cooling intensity of6.72 �C in Addis Ababa (Ethiopia) and concluded that the coolingeffect of urban green space is mainly determined by species group,canopy cover, size and the spatial design of parks. Moreover,Ivajn�si�c, Kaligari�c, and �Ziberna (2014) concluded that themaximumcooling distance of urban green space is rather small and it istherefore even more important in small cities where dispersedgreen areas could neutralize elevated air temperatures caused bythe UHI effect.

When public space is limited, other greening strategies aresuggested. Covering roofs or façades with plants might reduce thethermal load resulting in cooling of the urban environment(Campiotti et al. 2011; Wong & Lau, 2013). Green roofs and façadesrepresent a sustainable technology for improving both energy ef-ficiency of buildings, and thermal insulation in winter. Thoughattractive, their cooling effects have not been extensively studied(Bowler et al. 2010). Street tree planting in residential areas mightalso help in reducing UHI effect. An aggregate number of trees havean important impact on the thermal balance through shading andplant-transpiration. In this case, it is suggested to prioritize treesover grasses because they provide shade and require relativelylimited irrigation (House-Peters & Chang, 2011; Rosenzweig,Solecki, & Slosberg, 2006). In spite of the recent studies on theoverall cooling effects of residential greening, its local effectcompared to that of public parks has yet to be studied.

In this study, we investigated thermal patterns in Tel Aviv, Israel.Specifically, we compared the cooling effect of green residentialareas (i.e., buildings with private gardens and street trees) with thatof public parks within the neighborhood space. We used satellitedata to investigate the UHI effect in Tel Aviv, Israel. Land surfacetemperature and vegetation cover at 20 sites representing fourdifferent urban land uses were compared. High-temporal low-spatial resolution data were combined with low-temporal high-

spatial resolution using two different satellites to produce detailedtime series maps of the study area. The aim was to estimate andcompare the cooling effects of green residential areas and publicparks, in the neighborhood space. This paper seeks to answer twoquestions: (a) what are the thermal effects of different land uses ontheir surrounding in the metropolitan city of Tel Aviv? And (b) howcan the UHI effect be mitigated in a dense urban fabric? Answeringthese questions may help urban planners mitigate the UHI effect.By analyzing sites that display different types of land-uses we showthat there is a quantitative relation between the surface tempera-ture and physical features that can be modeled.

Data and methods

Study area

Tel Aviv, the core of Israel'smajormetropolitan area, is a dynamicurban area located in the most developed part of Israel. Althoughhaving an overall compact land-use structure, it includes areas withinefficient land use causing debate on its planning policy. Tel Aviv isthe largest city in the metropolitan area (52 km2) located along theeast coast of the Mediterranean Sea. Its 414,600 inhabitantscomprise almost 12% of the population in the entire metropolitanarea of Israel (CBS Israel, 2013). The city serves as the cultural heartand financial center of the country. It was developed as a relativelycompact city with mixed land use in which residential areasconverge with commercial, office, as well as industrial areas.

The climate is Mediterranean, characterized by subtropical hotand dry summers with extreme humid conditions during thesummer (30 �C and 83% relative humidity (Cohen, Potchter, &Matzarakis, 2013)). In the summer months (JuneeAugust), thesetemperature and humidity regimes are highly persistent (Saaroni&Ziv, 2000). Such conditions are commonly defined as dis-comfortable for humans (McGregor & Nieuwolt, 1998). Hot andhumid conditions in summer result from the combined effects ofthe Mediterranean Sea and urbanization (Potchter et al. 2006;Saaroni & Ziv, 2003). In summer, Tel Aviv develops an UHI effectat both midday and at night (Bitan, Noy, & Turk, 1992).

We selected Tel Aviv for two main reasons: (1) the nature of theurban fabric in its diverse neighborhoods, which enabled a com-parison between areas on a relatively small scale; and (2) its po-tential increase of thermal risk over the years due to a generalpolicy of residing in high densities as a principle guideline for thecity development (Tel Aviv Municipality, 2006).

Selected sites

In order to examine how UHI effect is expressed in differentlocal urban design, 20 sites comprising four urban land uses wereselected (Fig. 1):

(i) Green residential: Five sites of residential buildings withhigh vegetation of dense canopied trees and grasses, instreets and private gardens.

(ii) Residential: Five sites of residential buildings with little,usually low, vegetation. Vegetation consists mainly of shrubsor single trees.

(iii) Public parks: Five sites of small to medium sized green parkswith trees with/without grasses, within the neighborhoodspace.

(iv) Industrial: Five sites of dense building for industrial purposeswith no vegetation.

Fig. 1 and Table 1 present the location, view and description ofthe sites. On average there are three floors per building in the built

Fig. 1. Tel Aviv, the major metropolitan city of Israel. The 20 sites of this study that comprised four different urban land uses are presented in the map (see details in Table 1). Streetand closer aerial view of selected sites are also shown.

O. Rotem-Mindali et al. / Applied Geography 56 (2015) 145e153 147

sites (excluding public park sites). All potential sites were mappedfirst according to their size for each type of land-use. All sites wereselected far from the sea (~3 km), as a conditional criteria using GISsoftware. This was conducted to diminish the effect of sea breeze onsurface temperature (Lensky & Dayan, 2012), as sea breeze reducesthermal effect, decreasing its effect with increasing distance(Muller et al. 2013). The final sites were selected with expertknowledge to determine how well the sites fit the land-usesdefinitions.

To further understand the thermal pattern over the entire city,we divided it into five main urban land use types according to in-formation from the Tel Aviv municipality (Fig. 2a). As discussedearlier, Tel Aviv is a relatively compact city with mixed land-uses;however, in general it can be segmented into five main areas asillustrated in Fig. 2a. The metropolitan CBD and employment areasare located along the historical industrial belt and Ayalon highway,as well as a relatively new employment area located on the easternedge of Yarkon Park e a 3.5 square kilometer area comprising themain green urban area of Tel Aviv. The park is well connected to TelAviv's northern residential neighborhoods. These neighborhoodsare considered suburban residential neighborhoods with relativelylow connectivity to the Tel Aviv's central urban fabric. Commercialland use is somewhat dispersed in the city, including two mainshopping malls, while main shopping streets are located in 'central'

Tel Aviv. A main residential area in Tel Aviv is the ‘central’ Tel Avivcomplex refers to the enclosed area between the MediterraneanSea to the west, and the Ayalon Highway to the east. It includeshistoric residential areas, which were originally planned by PatrickGeddes as a modern Jewish garden city. The idea was to creategreen public spaces within residential blocks in the form of parksand squares. However, due to the growing needs and constraints oftime, the planned area development never fully materialized(Hysler Rubin, 2013).

Satellite data

Estimating UHI effect requires high spatial and temporal reso-lution data due to its high variability in time and space. Satelliteorbits and instrumentation yield a trade-off between their spatialand temporal resolutions, i.e., high spatial resolution at low tem-poral resolution and vice versa. Two of the main satellites in NASA'searth observing system are Landsat and Terra satellites. Landsatsatellites have been operating since the 1970s with high spatialresolution of 30-m in the spectral, and 60-m in the thermal bands,with a low repetition time of 16 days. Terra is a sun synchronoussatellite that was launched in 2000. It passes over any given point ofEarth's surface at around 10:30 a.m. and 10:30 p.m. local mean solartime, providing products such as land surface temperature (LST)

Table 1General description of the selected sites for this study comprising four urban landuses in Tel Aviv metropolitan (Green residential, residential, public park and in-dustrial). Veg. is the total evergreen vegetation cover during the summer (June-eAugust), and was calculated from the mean NDVI using Eq. (1) (see in Section 2.5).

Site Land use Description Total area Built area Veg.

(ha) (ha) (%) (%)

1 Green residential Apartment buildings 17.2 3.6 21 462 1e2 floors buildings 14.5 5.1 35 353 Apartment buildings 16.3 4.5 28 354 Apartment buildings 12.1 3.8 26 305 Apartment buildings 27.2 7.2 39 386 Residential Apartment buildings 7.9 3.3 45 137 Apartment buildings 21.3 6.2 39 88 Apartment buildings 9.5 3.2 45 139 Apartment buildings 3.3 1.4 42 1110 Apartment buildings 11.9 4.5 38 1411 Public park Park (near industrial) 2.6 0.1 4 2612 Park (near paved road) 12.8 0 0 4013 Park (near paved road) 9.8 0 0 4114 Park (near paved road) 42.4 0 0 4415 Park (near paved road) 13.6 0 0 2116 Industrial Industrial area 19.4 9.4 49 517 Commercial area 19.6 11.3 58 018 Commercial and offices 14.6 7.9 54 019 Commercial and offices 30.7 20.4 66 120 Industrial and offices 10.2 6.3 61 2

Fig. 2. Maps of Tel Aviv's (a) main urban land uses, and (b) mean surface temperature (LST, �Cthe Ayalon Highway (the main highway that crosses the city), and the Yarkon Park (the lar

O. Rotem-Mindali et al. / Applied Geography 56 (2015) 145e153148

and Normalized Difference Vegetation Index (NDVI), which is aspectral measure of vegetation. We used the LST and NDVI productsof these satellites, because they are freely available online withrelatively high temporal and spatial quality, which makes themvery reliable for such analysis.

The 1-km night LST (MOD11A2) and 250-m NDVI (MOD13Q1)products of theMODerate resolution Imaging Spectrometer (MODIS)onboard Terra (https://lpdaac.usgs.gov/products/modis_products_table/) were used to generate a 10-year time series over Tel Aviv.Temporal Fourier analysiswas applied to get the long-termmean LSTand NDVI during the summer (JuneeAugust) on a pixel basis, asdescribed in Lensky and Dayan (2011). Landsat thermal and spectraldata (http://earthexplorer.usgs.gov/) were calibrated according tothe USGS guide, and corrected for atmospheric absorption.

Downscaling procedure and LST spatial profiles

To downscale MODIS spatial resolution to a 30 m Landsat res-olution, we applied the pixel block intensity modulation techniqueproposed by Guo and Moore (1998):

LSTMODIS30 ¼ LSTMODIS

930 $LSTLandsat30

LSTLandsat930

where LSTMODIS30 is the 1-km MODIS LST downscaled to the 30-m

Landsat resolution; LSTMODIS930 is the 1-km MODIS LST data in its

) in night summer months (JuneeAugust). The 20 selected sites for this study (Table 1),gest park in Tel Aviv), are indicated in both maps.

O. Rotem-Mindali et al. / Applied Geography 56 (2015) 145e153 149

original resolution (925m) resampled to 930m; LSTLandsat30 is the LST

from Landsat channel 6 (60 m) resampled to 30 m; and LSTLandsat930 is

the mean Landsat LST over the MODIS LST pixel resolution (930 m).The same procedure was used to downscale the 250-m MODIS

NDVI data to 30-m:

NDVIMODIS30 ¼ NDVIMODIS

270 $NDVILandsat30

NDVILandsat270

where NDVIMODIS30 is the 250-m MODIS NDVI downscaled to the

30 m Landsat resolution; NDVIMODIS270 is the 250-mMODIS NDVI data

in its original resolution (231 m) resampled to 270 m; NDVILandsat30 isthe NDVI from the 30 m Landsat channels; and NDVILandsat270 is themean Landsat NDVI over the MODIS NDVI pixel resolution (270 m).

Then, LSTMODIS30 and NDVIMODIS

30 were used to calculate night LSTand the mean NDVI during the summer (JuneeAugust). We usednight summer temperatures because radiative cooling during thistime was seen to accentuate the spatial variability in LST (Lensky &Dayan, 2011).

To estimate the spatial thermal effect of green residential andsmall-medium public park sites we measured LST at 30 m intervals(Landsat pixel size) in three directions. We excluded the coastlinedirection from the analysis due to the sea breeze cooling effect(Muller et al. 2013). Directions were selected using Google earthmap (Google earth©) to assess that there was no significant landuse change between the site and the surrounding area. After 90 mthere was a significant change in land use at several sites; thus wedid not measure LST after that distance. ArcGIS (ESRI) was used tocalculate the spatial thermal profile and to produce contour mapsof LST.

Estimation of vegetation cover from NDVI

The Normalized Difference Vegetation Index (NDVI) is a goodsurrogate for vegetation cover (Huete et al. 2002) and biomassproduction (Helman, Lensky, Mussery, & Leu, 2014; Helman,Mussery, Lensky, & Leu, 2014). It is defined as (Rouse, Haas,Deering, Schell, & Harlan, 1974):

NDVI ¼ R0:8 � R0:6R0:8 þ R0:6

where R0.6 and R0.8 are the reflectance at the red (0.6 mm) and nearinfra-red (0.8 mm) bands, respectively. It is a unitless index,normalized to get values between �1 and 1, with a full vegetationcover approaching 1.

In Mediterranean-climate regions the NDVI during the summercomesmainly from the evergreenwoody vegetation. The evergreenvegetation cover can be estimated from NDVI if assuming homog-enous distribution of the vegetation species (Glenn, Huete, Nagler,& Nelson, 2008). Because this assumption is valid for our studysites, the percentage of vegetation cover (VC) could be calculated:

VC ¼ 100%$NDVIFull � NDVI

NDVIFull � NDVIGround[1]

where NDVI is the value to be converted to vegetation cover (%),NDVIFull and NDVIGround are NDVI values of full vegetative cover (forevergreen vegetation: NDVI ¼ 0.65) and bare ground (NDVI ¼ 0.04,which was the lowest value in industrial sites), respectively.

Statistical analyses

We used linear regression to assess relationships betweenmeanNDVI and LST, and between built areas and LST for the 20 sites.

Good correlation is considered when Pearson's (R) coefficient ap-proaches 1. To estimate the significance of the correlation, the two-tailed probability of a Pearson correlation coefficient was calculated(Cohen et al. 2013). The two-tailed student t-test was applied to testif the mean values of green residential and public park sites weresignificant. The one-way Anova was used to test significant differ-ence between the four land uses with a Bonferroni correction formultiple comparisons analysis. Data analysis was carried out usingstatistical software: Microsoft Office Excel 2011.

Results

Thermal patterns of Tel Aviv major urban land uses

Fig. 2b shows the thermal patterns of the major urban land usesin Tel Aviv during the summer as the mean night LST over ten years(2000e2010). Two areas with opposite thermal load can be noticed.

First, the metropolitan central business district (CBD) with thehighest LST: The CBD is a historic industrial strip located in thecentral area of Tel Aviv. Formerly, the industrial sector wasconcentrated in the southern and eastern part of Tel Aviv extendingnorth through the years toward the Ayalon Highway (the mainhighway that crosses the city). Presently, the CBD is located alongthe Ayalon Highway and is characterized by a dense built up areaincluding residential and nonresidential land uses. The thermalstress in this industrial strip is relatively high in comparison to otherlocations in Tel Aviv (Fig. 2b).

The second, with the lowest LST, is the Yarkon Park. Betweenthese two thermal extremes is the so-called ‘central’ Tel Avivcomplex.

Surface temperatures and NDVI in selected sites

Fig. 3a shows the cooling effect of vegetation in the sites fromFig. 1, in terms of mean summer NDVI and night LST. The rela-tionship observed in Fig. 3a is exponential, thus beyond a certainLST threshold, the cooling effect of vegetation is attenuated.

Green residential sites had a lower mean LST than public parks by0.5 �C (p < 0.05, using two-tailed student t-test), with an average of21.7 �C compared to 22.3 �C. This was in spite of a comparable vege-tation cover with mean NDVI of 0.25 and 0.26 (p > 0.5) in green res-idential and public park sites, respectively. Public parks had largespatial variability in vegetation cover expressed in large standarddeviations from the average NDVI (horizontal error bars in Fig. 3a).Thiswas also displayed in LST variability (vertical error bars in Fig. 3a).

Residential sites with low vegetation cover had relatively highLST (mean ¼ 22.8 �C), while industrial sites had the highest LST(mean ¼ 24.4 �C). Fig. 3b shows that LST was also highly, butpositively related (Pearson's (R) ¼ 0.89, p < 0.01) to built areas (%from total site area). The only exception was the public park sitesthat had higher LST than expected (indicated in Fig. 3b). There wasno correlation between the area and LST (p > 0.5).

The vegetation (% cover) to free space (% from total site area)cover ratio is shown in Fig. 4. Green residential sites had a signifi-cantly higher vegetation to free space ratio, with more than half ofthe free space covered by vegetation. Public parks had less than 35%of their free space covered with vegetation, which was also nearlytheir total area (see Table 1). Residential and industrial sites had thelowest vegetation to free space ratio, with 20% and <1% from theirfree space covered by vegetation, respectively.

Spatial thermal profiles around ‘green’ sites

LST generally decreased with distance when moving from thesite toward the surrounding areas, in all green residential and

a b

Fig. 3. The relationship between surface temperature (LST, �C) and (a) NDVI, a surrogate for vegetation cover, and (b) built area, in the 20 sites comprising different urban land uses(Table 1). The thermal effects of vegetation and built area are significant (P < 0.01). Error bars denote spatial variability within the site area as ±1s.

O. Rotem-Mindali et al. / Applied Geography 56 (2015) 145e153150

public park sites (Table 2). Fig. 5 shows the increase in LST from theborder of these two types of land use, at intervals of 30 m. The y-axis in Fig. 5 is the difference between the LST at distance x and atdistance x-30 m (DLST), while the x-axis is the distance x of the LSTmeasurement. Public parks have a strong cooling effect in the parkarea, but amuch smaller effect on their proximate surrounding area(i.e., high DLST after 30 m). Green residential areas (in the webversion), on the other hand, had a much smaller effect than that ofthe public park, but a more continuous one with a moderate in-crease in DLST with distance. The spatial patterns of LST are shownin Fig. 6 as LST contours with the mean NDVI also presented in thebackground.

Discussion

Efficient ‘green’ use of private gardens in residential area mitigatethermal load better than public gardens

Urban land uses with low vegetative cover were more affectedby the UHI effect in Tel Aviv. Differences in LST were up to 13 �Cwithin the city area (Fig. 2b). The role of vegetation in diminishingUHI was obvious in the case of Tel Aviv, where the Yarkon Park, thelargest in the city, had the lowest LST (average of 18.5 �C). This

Fig. 4. The mean ratio between vegetation cover and free space (in %) in the four landuse types (Table 1). Error bars denote standard error, with different letters indicatingsignificant difference between land use using one-way Anova test with a ¼ 0.1 afterBonferroni correction for multiple comparisons.

reinforces the literature indicating that large parks have a moreprominent cooling effect. On a local scale, vegetation (NDVI)attenuated LST in four different urban land uses, bolstering the roleof vegetation in mitigating UHI effect (Fig. 3a). A significant nega-tive correlation between NDVI and LST was demonstrated, as pre-viously reported also by Mackey, Lee, and Smith (2012). Thisresearch also demonstrated that a built area affects the thermalload, with highest LST measured in industrial sites where the builtarea was higher than 49% (Fig 3b). This emphasizes the contribu-tion of surface cover, characterized by high albedo, to the high LST.

A thought-provoking result revealed that public parks (small-medium sized), which do not contain built areas, displayed higherLST than expected (Fig. 3b). Apparently, it could be expected thatsuch public parks would have a lower LST compared to green res-idential areas. However, a closer look into the related literatureshows that research on public parks often concentrates on mediumand large parks (100e500 ha), while smallemedium parks(3e50 ha) have been less studied. Indeed, studies point to therelationship between the park's size and its cooling effect in me-dium to large parks (Alcoforado et al. 2009; Andrade & Vieira,2007; Eliasson and Upmanis 2000; Jauregui, 1990; Shashua-Bar &Hoffman, 2000), but the effect of smaller scale parks is not asobvious (Andrade & Vieira, 2007). Therefore, it is possible that insmall-medium public parks, inhibiting factors may exist thatreduce their cooling effect.

Table 2The LST measured at 30 m intervals from the site in counter direction to thecoastline, in ‘green residential’ and ‘public parks’ land uses (See Section 2.4 forexplanation).

Land use Site LST (�C)

0 m 30 m 60 m 90 m

Green residential 1 20.9 21.2 21.4 21.62 21.2 21.5 21.6 21.93 21.7 22.0 22.2 22.54 22.6 22.7 22.8 23.05 22.3 21.9 22.2 22.8

Public park 11 21.9 21.9 22.5 22.612 21.8 22.1 22.4 22.613 22.2 22.4 23.1 23.714 23.1 24.0 23.9 24.015 22.6 23.9 24.7 24.6

Fig. 5. Change in surface temperature (LST, �C) versus the distance from the sites'border for residential with vegetation and public park sites. Y-axis is the differencebetween the temperature measured at distance x and the measured at distance x-30 m. The distance 0 m is the LST measured at the site. Error bars denote ±1s.

O. Rotem-Mindali et al. / Applied Geography 56 (2015) 145e153 151

For example, Fig. 4 indicates that a factor that could explain thisis the low vegetative cover in small public parks. The ‘green’ resi-dential neighborhoods used their free space more efficiently forvegetation cover, with dense tree and shrub plantings in private

Fig. 6. Spatial pattern of surface temperature presented as LST contours (�C) around two res

background.

gardens and facades. The status of the vegetation (i.e., wateravailability), could also contribute to the low NDVI in public parks,diminishing their cooling effect. More intensive irrigation in privategardens, relative to public parks, could result in higher NDVI,without necessarily indicating higher vegetation cover. In suchcases, irrigated vegetation would mitigate thermal load throughtranspiration. Also, unlike previously reports (Alcoforado et al.2009; Eliasson and Upmanis 2000), the size of the park did notcontribute to the parks' cooling effect in small-medium size parks.

Location of public parks affects their spatial thermal load

Residential areas with vegetation also differed from neighbor-hood public parks in their spatial thermal effect. The change intemperature from small gardens integrated in residential land usetoward the surrounding areas was moderate compared to that ofpublic gardens (Figs. 5 and 6). This could result from the location ofthe public urban parks. Public parks are usually located near pavedroads with high thermal load (see Table 1 and Fig. 6). Occasionally,public parks are also excluded from a continuous green surround-ing. The outcome may be a significant increase in temperaturewhen moving away from the public park area towards the sur-roundings. For example, major roads surround site 14, while site 15is located next to a concrete parking lot. In both sites, the sur-rounding land-cover causes a relatively high thermal load.

idential with vegetation and two public park sites. The mean NDVI is also shown in the

O. Rotem-Mindali et al. / Applied Geography 56 (2015) 145e153152

On the other hand, we find that generally, private gardens arelocated within continuously similar land use coverage. Becausegreen residential areas usually indicate higher socio-economicstatus, this is most likely to affect the surrounding area in termsof vegetation cover (Kamphuis et al. 2010). Consequently, thetemperature is also expected to be relatively low in such areas andthe spatial effect of green residential areas, more continuous. Anaggregate effect of green residential areas may result in anextended cooling effect on a wider surrounding area. Eventually, itcould be compared to a park cooling island formation and may beconsidered as green residential cooling island (GRCI).

Nevertheless, it is important to stress that it is not accurate toconclude that small-medium public parks are not essential tomitigate thermal load. On the contrary, they have an important roleespecially on areas with high albedo. In addition, keeping urbanopen space is an important means that positively influences thephysical health of the population (Zhou & Rana, 2012).

Finally, in Tel Aviv, one may notice the high NDVI to the north ofthe Yarkon Park and in the central (historic) area of Tel Aviv, whilethe southern area of Tel Aviv demonstrates lower NDVI (as well ashigher LST (Fig. 2a, b)). Hence, it is central to highlight this disparity,not only from socio-economic aspect, but also from the physicalaspect of land use coverage. Such disparity is displayed in terms oflower thermal load, affecting physical aspects of the quality of life.

Summary and implications for urban planning

This research analyses the effect of different land uses on surfacetemperature in the Tel Aviv metropolitan area. Four urban land usetypes including: residential area with vegetation, residentialwithout vegetation (or low vegetation cover), public parks andindustrial areas were analyzed. The selected public parks weresmall to medium size neighborhood parks (from � 3 to � 40 ha)and not largemetropolitan parks, such as the Yarkon Park that has agreater cooling effect.

There was a similarity, in terms of vegetation cover and tem-perature, between ‘green’ residential sites and neighborhood publicparks. However, public park space was less efficiently used forvegetation cover than private gardens in residential sites. As shownearlier, vegetation was denser in private gardens (vegetation cover/free space). A possible explanation is that irrigation in privategardens was better than in public parks, affecting plant transpira-tion and reducing thermal load more efficiently. Also, the locationof the public parks, which is usually near roads and parking lotsreduce their cooling effect.

Where vegetation coverage for mitigating UHI is the target,planners should characterize the particular purposes, i.e., creating'green islets' to lower local thermal load. In a dense urban fabric thiscould be done not only by adding small to medium public parks,which are important for leisure activities, human interaction, aswell as for physical health, but also by creating continuous greenresidential land use. By forming green residential neighborhoods,green residential cooling islands (GRCI) are created while maxi-mizing the use of free spaces and maintaining a high quality of life.Hence, the aspects discussed in this study may have key implica-tions for urban planners seeking to mitigate the UHI effect underthe limitation of existing dense urban layout.

Acknowledgment

The authors thank Dr. Ilan Levy for his valuable contribution tothis research. The authors also thank the anonymous reviewers fortheir valuable and constructive suggestions.

References

Alcoforado, M.-J., Andrade, H., Lopes, A., & Vasconcelos, J. (2009). Application ofclimatic guidelines to urban planning: the example of Lisbon (Portugal).Landscape and Urban Planning, 90(1e2), 56e65.

Andrade, H., & Vieira, R. (2007). A climatic study of an urban green space: theGulbenkian Park in Lisbon (Portugal). Finisterra: Revista Portuguesa de Geografia,42(84), 27e46.

Bitan, A., Noy, L., & Turk, R. (1992). The impact of the seashore on the climate of Tel-Aviv. Scientific Reports of the Institute for Meteorology and Climate Research of theKarlsruhe University, 16, 147e160.

Bowler, D. E., Buyung-Ali, L., Knight, T. M., & Pullin, A. S. (2010). Urban greening tocool towns and cities: a systematic review of the empirical evidence. Landscapeand Urban Planning, 97(3), 147e155.

Campiotti, C., Bibbiani, C., Alonzo, G., Giagnacovo, G., Ragona, R., & Viola, C. (2011).Green roofs and facades agriculture (GRF) for supporting building energy effi-ciency. Journal of Sustainable Energy, 2(3).

CBS Israel. (2013). Statistical abstracts of Israel Jerusalem, Israel.Chang, C.-R., Li, M.-H., & Chang, S.-D. (2007). A preliminary study on the local cool-

island intensity of Taipei city parks. Landscape and Urban Planning, 80(4),386e395.

Cohen, P., Potchter, O., & Matzarakis, A. (2012). Daily and seasonal climatic condi-tions of green urban open spaces in the Mediterranean climate and their impacton human comfort. Building and Environment, 51, 285e295.

Cohen, P., Potchter, O., & Matzarakis, A. (2013). Human thermal perception ofCoastal Mediterranean outdoor urban environments. Applied Geography, 37,1e10.

Eliasson, I. R., & Upmanis, H. (2000). Nocturnal airflow from urban parks-implications for city ventilation. Theoretical and Applied Climatology, 66(1e2),95e107.

Feyisa, G. L., Dons, K., & Meilby, H. (2014). Efficiency of parks in mitigating urbanheat island effect: an example from Addis Ababa. Landscape And Urban Planning,123, 87e95.

Gill, S., Handley, J., Ennos, A., & Pauleit, S. (2007). Adapting cities for climate change:the role of the green infrastructure. Built Environment (1978-), 115e133.

Glenn, E. P., Huete, A. R., Nagler, P. L., & Nelson, S. G. (2008). Relationship betweenremotely-sensed vegetation indices, canopy attributes and plant physiologicalprocesses: what vegetation indices can and cannot tell us about the landscape.Sensors, 8(4), 2136e2160.

Guo, L. J., & Moore, J. M. (1998). Pixel block intensity modulation: adding spatialdetail to TM band 6 thermal imagery. International Journal of Remote Sensing,19(3), 2477e2491.

Helman, D., Lensky, I. M., Mussery, A., & Leu, S. (2014). Rehabilitating degradeddrylands by creating woodland islets: assessing long-term effects on above-ground productivity and soil fertility. Agricultural and Forest Meteorology, 195,52e60.

Helman, D., Mussery, A., Lensky, I., & Leu, S. (2014). Detecting changes in biomassproductivity in a different land management regimes in drylands usingsatellite-derived vegetation index. Soil Use and Management, 30(1), 32e39.

House-Peters, L. A., & Chang, H. (2011). Modeling the impact of land use andclimate change on neighborhood-scale evaporation and nighttime cooling: asurface energy balance approach. Landscape And Urban Planning, 103(2),139e155.

Huete, A., Didan, K., Miura, T., Rodriguez, E. P., Gao, X., & Ferreira, L. G. (2002).Overview of the radiometric and biophysical performance of the MODISvegetation indices. Remote Sensing of Environment, 83, 195e213.

Hysler Rubin, N. (2013). The celebration, condemnation and reinterpretation of theGeddes plan, 1925: the dynamic planning history of Tel Aviv. Urban History,40(01), 114e135.

Ivajn�si�c, D., Kaligari�c, M., & �Ziberna, I. (2014). Geographically weighted regression ofthe urban heat island of a small city. Applied Geography, 53, 341e353.

Jauregui, E. (1990). Influence of a large urban park on temperature and convectiveprecipitation in a tropical city. Energy and Buildings, 15(3e4), 457e463.

Johnson, D. P., Stanforth, A., Lulla, V., & Luber, G. (2012). Developing an appliedextreme heat vulnerability index utilizing socioeconomic and environmentaldata. Applied Geography, 35(1), 23e31.

Kamphuis, C., Mackenbach, J. P., Giskes, K., Huisman, M., Brug, J., & Van Lenthe, F. J.(2010). Why do poor people perceive poor neighbourhoods? the role ofobjective neighbourhood features and psychosocial factors. Health & Place,16(4), 744e754.

Kleerekoper, L., van Esch, M., & Salcedo, T. B. (2012). How to make a city climate-proof, addressing the urban heat island effect. Resources Conservation andRecycling, 64, 30e38.

Kovats, R. S., & Hajat, S. (2008). Heat stress and public health: a critical review.Annual Review of Public Health, 29, 41e55.

Lensky, I. M., & Dayan, U. (2011). Detection of Finescale climatic features fromsatellites and implications for agricultural planning. Bulletin of the AmericanMeteorological Society, 92(9), 1131e1136.

Lensky, I., & Dayan, U. (2012). Continuous detection and characterization of the SeaBreeze in clear sky conditions using Meteosat Second Generation. AtmosphericChemistry and Physics, 12(14), 6505e6513.

Lin, Y., Liu, A., Ma, E., Li, X., & Shi, Q. (2013). Impacts of future urban expansion onregional climate in the Northeast Megalopolis, USA. Advances in Meteorology,2013.

O. Rotem-Mindali et al. / Applied Geography 56 (2015) 145e153 153

Mackey, C. W., Lee, X., & Smith, R. B. (2012). Remotely sensing the cooling effects ofcity scale efforts to reduce urban heat island. Building and Environment, 49,348e358.

McGregor, G. R., & Nieuwolt, S. (1998). Tropical climatology. Chichester: John Wiley& Sons.

Mirzaei, P. A., & Haghighat, F. (2010). Approaches to study urban heat islande-abilities and limitations. Building and Environment, 45(10), 2192e2201.

Muller, N., Kuttler, W., & Barlag, A.-B. (2013). Counteracting urban climate change:adaptation measures and their effect on thermal comfort. Theoretical andApplied Climatology, 1e15.

Oke, T. R. (1995). The heat island of the urban boundary layer: characteristics,causes and effects. NATO ASI Series E Applied Sciences-Advanced Study Institute,277, 81e108.

Oke, T. R., Crowther, J., McNaughton, K., Monteith, J., & Gardiner, B. (1989). Themicrometeorology of the urban forest [and discussion]. Philosophical Trans-actions of the Royal Society of London. B, Biological Sciences, 324(1223), 335e349.

Oliveira, S., Andrade, H., & Vaz, T. (2011). The cooling effect of green spaces as acontribution to the mitigation of urban heat: a case study in Lisbon. Buildingand Environment, 46(11), 2186e2194.

Papangelis, G., Tombrou, M., Dandou, A., & Kontos, T. (2012). An urban “œgreenplanning” approach utilizing the Weather Research and Forecasting (WRF)modeling system. A case study of Athens, Greece. Landscape And Urban Plan-ning, 105(1e2), 174e183.

Petralli, M., Massetti, L., Brandani, G., & Orlandini, S. (2014). Urban planning in-dicators: useful tools to measure the effect of urbanization and vegetation onsummer air temperatures. International Journal of Climatology, 34(4),1236e1244.

Potchter, O., Cohen, P., & Bitan, A. (2006). Climatic behavior of various urban parksduring hot and humid summer in the Mediterranean city of Tel Aviv, Israel.International Journal of Climatology, 26(12), 1695e1711.

Ren, Z., He, X., Zheng, H., Zhang, D., Yu, X., Shen, G., et al. (2013). Estimation of therelationship between urban park characteristics and park cool island intensityby remote sensing data and field measurement. Forests, 4(4), 868e886.

Rogan, J., Ziemer, M., Martin, D., Ratick, S., Cuba, N., & DeLauer, V. (2013). The impactof tree cover loss on land surface temperature: a case study of central Massa-chusetts using Landsat Thematic Mapper thermal data. Applied Geography, 45,49e57.

Rosenzweig, C., Solecki, W., & Slosberg, R. (2006). Mitigating New York City's heatisland with urban forestry, living roofs, and light surfaces. A report to the NewYork State Energy Research and Development Authority.

Rouse, J. W. J., Haas, H. R., Deering, D. W., Schell, J. A., & Harlan, J. C. (1974). Moni-toring the vernal advancement and retrogradation (green wave effect) of naturalvegetation. NASA/GSFC Type III Final Rep, Greenbelt, Md: 371.

Saaroni, H., & Ziv, B. (2000). Summer rain episodes in a Mediterranean climate, thecase of Israel: climatological-dynamical analysis. International Journal ofClimatology, 20(2), 191e209.

Saaroni, H., & Ziv, B. (2003). The impact of a small lake on heat stress in a Medi-terranean urban park: the case of Tel Aviv, Israel. International Journal ofBiometeorology, 47(3), 156e165.

Shashua-Bar, L., & Hoffman, M. E. (2000). Vegetation as a climatic component in thedesign of an urban street e an empirical model for predicting the cooling effectof urban green areas with trees. Energy and Buildings, 31(3), 221e235.

Spronken-Smith, R., & Oke, T. (1998). The thermal regime of urban parks in twocities with different summer climates. International Journal of Remote Sensing,19(11), 2085e2104.

Stone, B., Hess, J. J., & Frumkin, H. (2010). Urban form and extreme heat events: aresprawling cities more vulnerable to climate change than compact cities?Environmental Health Perspectives, 118, 1425e1428.

Tel Aviv Municipality. (2006). City, the strategic plan for Tel Aviv Yafo. Retrieved01.01.14 from http://www.tel-aviv.gov.il/eng/AboutTheCity/Pages/StrategicPlan.aspx.

Upmanis, H., Eliasson, I., & Lindqvist, S. (1998). The influence of green areas onnocturnal temperatures in a high latitude city (Goteborg, Sweden). InternationalJournal of Climatology, 18(6), 681e700.

Wong, J. K. W., & Lau, L. S.-K. (2013). From the ,urban heat islan's to the ,greenislands? A preliminary investigation into the potential of retrofitting greenroofs in Mongkok district of Hong Kong. Habitat International, 39(0), 25e35.

Yu, C., & Hien, W. N. (2006). Thermal benefits of city parks. Energy and Buildings,38(2), 105e120.

Zhou, X., & Rana, M. M. P. (2012). Social benefits of urban green space: a conceptualframework of valuation and accessibility measurements. Management of Envi-ronmental Quality: An International Journal, 23(2), 173e189.